Dr. @noriyuki_kojima will give a talk on #AI for language and communication at #GStar Summit 2026.
His session will explore how real-time speech AI can help more people participate across languages.
Let's be part of the talk: https://t.co/RgPHIjf2fw
For years, Vietnam has been part of the global tech supply chain.
The next question is more ambitious: can Vietnamese companies build #AI products and capabilities that compete globally?
Dr. Phong Nguyen, CTO of #FPT Corporation, will speak to this shift at #GStar Summit 2026.
Explore more: https://t.co/mHX1H6DOjM
Prof. @PoShenLoh will deliver a keynote at #GStar Summit 2026 on #AI’s disruption of #education, alongside a panel discussion on #Humanity.
Come listen, reflect, and explore: https://t.co/9JtMDBX24q
@marcwoohf will join #GStar Summit 2026 for a conversation on #AI industry and #ecosystem development.
Hear Marc Woo’s perspective: https://t.co/9JtMDBX24q
As foundation models, compute, and large-scale data concentrate in a few global tech hubs, emerging markets face a strategic question: how to adopt AI without becoming overly dependent on systems they do not govern.
At #GStarSummit, Jean DeSombre will bring a policy and governance lens to this challenge.
Curious to hear from the #AI, #policy, and #investment community: what should emerging markets prioritize first to build long-term AI agency?
LLMs are changing the misinformation threat model: false claims can now be paraphrased, contextualized, and scaled with few surface-level signals.
At GStar Summit 2026, Prof. @preslav_nakov will discuss AI safety, fact-checking, propaganda, media bias detection, and the trust layer of AI.
Join us at: https://t.co/mHX1H6DOjM
At GStar Summit 2026, Dr. @lmthang will add a research lens shaped by foundation models and frontier AI
Full program and updates: https://t.co/9JtMDBX24q
Dr. @edchi will join GStar Summit 2026 for a keynote on foundation models and a panel discussion on frontier AI.
Website for program details and ticket registration: https://t.co/9JtMDBX24q
Be part of the conversation!
‼️ LLM asks smaller model for help: 80% cheaper, better accuracy
Inspired by @NVIDIAAI 's "Small Language Models are the Future of Agentic AI", I built a framework where an LLM can ask a specialized Small Language Model (SLM) for assistance on 'easy' tasks. This is because SLMs can sufficiently perform such tasks, while costing much less than an LLM.
For simplicity + speed, I tested on math reasoning tasks (GSM8K, 500 Qs).
Results vs. LLM (alone)
• ✅ 80% fewer tokens
• ✅ 1.8% accuracy improvement
• ❌ 140% slower
☄️ Next Steps:
• Expand Routing: Route tasks based on SLM performance strengths.
• Optimize Latency: Improve SLM inference speed
• Expand Domains: Extend evaluation to more 'agentic' tasks (e.g. Deep Search)
I'd really appreciate feedback + ideas!
cc @newturing
(Github link below 👇)
Just had a really fun and insightful lecture with Professor Christopher Manning, one of the co-inventors of the GloVe word vectors!!
His life advice: “Jump into promising new things early (with both feet)”
He also commented on Yann LeCun’s “the future of AI is not LLMs”, saying that such statement greatly underestimates language and how much it represents intelligence
We talked about his research journey and interests, promising research areas, as well as communication w/ aliens 👽
He also talked about how important it is for researchers to have a deep technical expertise and understand, and recommends aspiring researchers to implement paper to build that understand
It was a great lecture @newturing, and thank you so much to Dr. @lmthang and the team for having invited Professor Manning!!
post-training lecture by @jon_lee0 who co-led the recent Gemini’s IMO gold medal at @GoogleDeepMind
felt surreal to be learning from the best as part of @newturing
Congrats @furst_fly! Special thanks to @chiennv2000 and @longphan3110 for designing this Flash Attention assignment. Check it out at https://t.co/LA5qhD63i4!
It felt like yesterday I was a 17yo at MIT Manipal who thought AI was more magic than math.
Today, I shipped a first-of-its-kind backprop implementation that combines FlashAttention + GQA + SWA + Attention Sinks for GPT‑OSS : fully open source.
Benchmarks: ~33× faster fine-tuning and ~138× more memory‑efficient vs naïve PyTorch, on the same hardware. Visuals attached.
HuggingFace has FlashAttention support, but nothing close to this training efficiency/practicality existed…until now.
Want to try it?
Repo link:https://t.co/rqDvlYwEEZ
run a finetune and tell me what breaks.
Huge thanks to @lmthang, @longphan3110 for putting this problem on my radar, And big shoutout to the @newturing 's GStar team including @JeffDean , @YiTayML and everyone who helped along the way.
I too really enjoyed the talk at the GenAI summit by @PoShenLoh. He has a really innovative way of getting high school kids to teach other kids math in an engaging manner ("Twitch live stream for math problems") while giving teachers immediate feedback.
https://t.co/eIJZDJ6qSL
Had a great weekend being part of the GenAI summit in Vietnam. 😄
Really enjoyed the impressive talks by @jeffdean, @quocleix@lmthang and @Diyi_Yang, along with the nice panel discussion that we had. 😄
Meanwhile, one major unexpected highlight of this trip for me personally was meeting @PoShenLoh and watching his talk on math, education, game theory and changing the world. This man is built different and cracked. Really such a breath of fresh air admist all the AI hype. 😄
The production quality of the summit was super serious and high quality 👌. The energy of the local crowd really makes me feel that Vietnam is really serious about AI. 🫡
This is my first time in Vietnam and it was amazing to be hosted by @lmthang. On the second day we visited some universities and left me a strong impression about the culture there. People there seem hungry for learning and science/math and that is amazing.
Enjoyed the trip and the great hospitality and thoughtfulness by the hosts and organizers. 👍
Nit: Vietnam needs more ACs though🥵🥵🥵
I'm excited to speak at the GenAI summit in Vietnam (and to visit Vietnam for the first time). Thanks to @lmthang and the other organizers for putting together a very interesting set of speakers including @quocleix , @Diyi_Yang , @YiTayML and others!